Live commercial buildIC Solar: AI Website & Growth System
An interactive B2B website with customer tools, AI guidance, and structured lead generation. Live at icsolar.rs.
View case studyWebsites, customer tools, automation, and internal software that connect AI with real business workflows - without removing human control.

If you already have customers and a process that works, but part of your business is still slow, manual, or limited by existing software, I can help improve it with AI and custom software.
The right solution might be an AI tool, an automation, a better interface, or a combination of all three.
Three selected builds show how I approach customer-facing software, production automation, and internal business systems.
Live commercial buildAn interactive B2B website with customer tools, AI guidance, and structured lead generation. Live at icsolar.rs.
View case studyResearch, deduplication, human approval, bounded daily recovery, and weekly discovery in one controlled pipeline.
View case study
Custom business prototypeA reviewable sales workflow that connects partner leads, ERP/customer records, lead scoring, and human-approved outreach.
View case studyWebsites that do more than present a company - interactive calculators, AI assistants, qualification flows, customer tools, and CRM handoff.
Research, classification, scoring, structured outputs, evaluator steps, and decision support around a real process.
Data cleanup, reply matching, status synchronization, customer context, and follow-up preparation across existing tools - with approval-gated n8n pipelines and scheduled recovery for production workflows.
Dashboards, client portals, growth intelligence, and operational workspaces for teams that have outgrown spreadsheets and disconnected software.
These counters describe what is publicly documented in this portfolio. Time and savings depend on the visitor's workflow, so the calculator below estimates their workload instead of claiming mine.
documented systems
automation case studies
selected homepage builds
anonymized production automations
I focus on the handoffs where information becomes scattered, decisions slow down, and people repeat the same operational steps every day.
Customer records, inboxes, spreadsheets, supplier research, and internal context rarely live in one reliable place.
Teams repeatedly sort, compare, score, summarize, and prepare the same work before a decision can move forward.
The site exists, but customers still have to call or email to get an estimate, an answer, or a clear next step.
A quick estimate of how repetitive and structured the process is, how much manual work it currently takes, and where human review still matters.
Not sure what counts as one repetition? Load a realistic example - every value stays editable.
Swipe to see more →
Two anonymized n8n systems show the reliability patterns behind my current production work.
Research, deduplication, human approval, bounded daily recovery, and weekly discovery in one controlled pipeline.
Running in productionMulti-mailbox monitoring, fallback matching, CRM status synchronization, and an overdue-only follow-up digest.
Practical design decisions around reliability, human control, and messy operational data.
The approval step often matters more than the AI model because it defines where risk stops.
Field note · 4 min02A bounded recovery path can be more reliable than making the live workflow increasingly complicated.
Field note · 4 min03A short fallback chain keeps CRM matching useful when email threading stops being reliable.
Field note · 3 min
I got into automation by watching capable people lose hours to work a machine should prepare: retyping data, chasing confirmations, cross-checking spreadsheets. My electrical engineering background taught me to think in inputs, constraints, and failure modes - and that is exactly how I build software.
What drives me is the moment a system quietly does the preparation and a person gets to make the real decision. So everything I ship stays small, inspectable, and under human control - because trust in a system is earned the same way as trust in a person: by being predictable when it matters.
Whether it's a website that should do more for customers, a repetitive workflow, or an internal process that has outgrown spreadsheets - start with how it works today, and I'll help you identify the smallest useful system to build.
Projects are scoped and priced by the problem: the scope, the integrations, the data and content involved, and how much custom software it takes - not by the hour.

Checking the portfolio context…